|Home | About | Journals | Submit | Contact Us | Français|
The control of normal cell growth is a balance between stimulatory and inhibitory signals. MYC is a pleiotropic transcription factor that both activates and represses a broad range of target genes and is indispensable for cell growth. While much is known about gene activation by MYC, there is no established mechanism for the majority of MYC repressed genes. We report that MYC transcriptionally activates the PTEN tumor suppressor in normal cells to inactivate the PI3K pathway, thus suppressing AKT activation. Suppression of AKT enhances the activity of the EZH2 histone methyltransferase, a subunit of the epigenetic repressor Polycomb Repressive Complex 2 (PRC2), while simultaneously stabilizing the protein. MYC mediated enhancement in EZH2 protein level and activity results in local and genome-wide elevation in the repressive H3K27me3 histone modification, leading to widespread gene repression including feedback autoregulation of the MYC gene itself. Depletion of either PTEN or EZH2 and inhibition of the PI3K/AKT pathway leads to gene derepression. Importantly, expression of a phospho-defective EZH2 mutant is sufficient to recapitulate nearly half of all MYC-mediated gene repression. We present a novel epigenetic model for MYC-mediated gene repression and propose that PTEN and MYC exist in homeostatic balance to control normal growth which is disrupted in cancer cells.
Cancer is driven by the activation of oncogenes and the loss of tumor suppressors. One tumor suppressor that is frequently inactivated in diverse cancers is the phosphatase and tensin homolog gene (PTEN) (reviewed in (1)). PTEN acts as a negative regulator of the PI3K pathway by converting PIP3 to PIP2, which in turn prevents activation of the Akt pathway downstream of PI3K. The PI3K/Akt pathway is an important component of cell signaling that regulates a myriad of biological processes including growth, proliferation and apoptosis (1). Cancer cells often exhibit mutations in the PI3K/Akt pathway, including loss of PTEN, ultimately resulting in activation of the pathway and its downstream effectors (2).
The MYC gene is the most frequently amplified gene in human cancer, and deregulated expression of MYC is a hallmark of 70% of all cancers (3). Myc is a well-established pleiotropic transcription factor and significant progress has been made in understanding the role of Myc as a transcriptional activator (4). In addition to activating target genes, Myc also represses an almost equal number of genes (3), and this repression is important for Myc mediated cell proliferation and transformation (5, 6). However, despite major advances in the Myc field, there is no uniform mechanism of repression that satisfactorily accounts for the majority of Myc repressed genes. In fact, the first Myc regulated gene ever described was the MYC gene itself, through what is generally considered an autoregulatory feedback loop (7, 8). Autoregulation is postulated to be important in fine-tuning the amount of Myc in a cell since small changes in Myc expression are sufficient to shift the balance from normal to aberrant growth. Notably, many cancer cells have lost the ability to autoregulate (9, 10), and more Myc is advantageous for growth and proliferation of cancer.
One pathway implicated in autoregulation of the MYC ortholog in Drosophila (dmyc) is the Polycomb group of proteins (11). The Polycomb Repressive Complex 2 (PRC2) epigenetically silences genes by trimethylating lysine 27 on histone H3 (H3K27me3), a function carried out by the methyltransferase Enhancer of Zeste 2 (Ezh2) which is an integral component of the complex (12). The Ezh2 gene is frequently amplified or overexpressed in many tumors and was described as an E2F responsive oncogene (13).
In this study, we show that Myc suppresses the PI3K/Akt pathway via transcriptional upregulation of the PTEN tumor suppressor. Significantly, suppression of Akt results in Ezh2-mediated gene repression in two mammalian systems, including autoregulation. Activation of Ezh2 is both necessary and sufficient to account for nearly half of all Myc repressed genes. We propose a general mechanism for Myc-mediated repression and autoregulation linked to an important tumor suppressor pathway.
c-myc−/− (HO 15.19) and Phoenix cells were maintained in DMEM supplemented with 10% fetal bovine serum. Immortalized mammary epithelial cells (IMECs) (14) were cultured in DMEM:F12 50:50 media supplemented with epidermal growth factor, insulin, hydrocortisone and 5% FBS. To generate stable cell lines, retroviral vectors were used to create polyclonal populations. For LY294002 (Cayman Chemical; #70920) treatments, cells were plated subconfluently 12–16 hours prior to treatment. Treatments lasted for 2 hours followed by RIPA lysis for immunoblotting. Antibody information used for immunoblotting and ChIPs can be found in Supplemental Information.
To generate the myc promoter-luciferase reporter, a 2.5 kb fragment containing the human MYC promoters (P1 and P2) and upstream regulatory elements was cloned into the pGL3 Basic vector. Ezh2/S21A and Ezh2/S21E mutants were generated using the QuikChange II Site Directed Mutagenesis Kit (Stratagene) as per manufacturers protocol. pUSE MYR-AKT was a gift from Dr. Lienhard (Dartmouth). Silencer Select Pre-designed siRNAs were obtained from Ambion (Applied Biosystems) and transfected using Lipofectamine RNAi Max (Invitrogen; 10nM siRNA). At the indicated time point, cells were harvested for protein or RNA. siRNA sequences are listed in Supplemental Information (Supplemental Table S2). Silencer Select Negative Control siRNA was used as a transfection control and to account for non-specific effects.
Total RNA was harvested from log phase cultures with Trizol (Invitrogen) and cDNA was synthesized using the Super Script kit from Invitrogen. Two-step real-time polymerase chain reaction (PCR) was performed using the SYBR Green Mix (BioRad) on a BioRad C1000 Thermal cycler. The expression of GAPDH or ACTIN was used for normalization. Primer sequences are available upon request.
Log phase cultures were fixed with 1% formaldehyde for 10 min and subjected to ChIP assays with minor modifications as previously described (15). Briefly, cells were lysed in Lysis Buffer (1% SDS, 10mM EDTA, 50mM TRIS pH8.1) and sonicated to generate DNA fragments between 200–1000bps. Cleared lysates were then diluted and incubated with the following antibodies overnight at 4°C. IPs were washed with RIPA buffer and precipitated DNA was recovered. Real-time PCR amplification was then performed using specific primers. Primer sequences are available upon request. The data is presented as % binding compared to input for each sample. All experiments were performed three to four times and error bars represent S.D.
Luciferase assays were performed with the Dual luciferase Reporter Kit (Promega). The pRL vector constitutively expressing Renilla luciferase was used to normalize for transfection efficiency. 2×105 IMECs were plated in 12 well dishes 24 hours prior to transfections. On the day of transfection, each well was transfected with pRL, pGL3 Basic (to asses basal reporter activity) or MYC promoter-GL3 and the indicated plasmids. 24 hours later, luciferase activity was measured using the Wallac 1450 MicroBeta TriLux system (Perkin Elmer). Experiments were carried out three times in triplicates and error bars represent S.D.
In 6 well plates, 10,000 cells (+/− siRNA treatment for 48 hours) were plated in 0.3% agar, layered over a 0.6% agar base layer. Wells were re-fed with 200ul complete media every other day. 14 days after plating, colonies were counted and imaged.
The PTEN tumor suppressor was previously shown to be a direct Myc target gene with an E-box that is occupied by Myc in vivo (16, 17). We analyzed PTEN expression and found that it is Myc-activated at both the protein and mRNA levels in human mammary epithelial cells (IMECs) and myc −/− rat fibroblast cells (Figures 1A and 1B). Therefore, we decided to explore the functional consequences of PTEN activation by Myc.
PTEN is a dual lipid and protein phosphatase that acts as a negative regulator of the PI3K/Akt pathway (1). Since PTEN dephosphorylates PIP3 to PIP2, thus attenuating Akt activation, we tested if PTEN activation had an effect on Akt signaling. Activation of Akt involves phosphorylation of Serine 473 (pS473) so we assessed the levels of pS473 in response to exogenous Myc expression. Notably, we found that Myc inhibits Akt S473 phosphorylation in both IMECs and myc−/− rat fibroblasts (Figure 1C). In IMECs, there was a basal level of pS473 in the parental/vector cells that becomes undetectable with Myc overexpression. In rat fibroblasts, there is no detectable pS473 in myc+/+ cells which express native levels of Myc. In contrast, myc−/− cells derived by genetic knockout have highly elevated pS473 which is subsequently suppressed to undetectable levels by reconstitution of Myc. Repression of Akt pS473 is not evident with a Myc mutant lacking the conserved Myc Box 2 transactivation domain (ΔMB2) (Figure 1C). MB2, an evolutionarily conserved motif in the Myc transactivation domain, is essential for most of Myc’s biological activities including oncogenic transformation, transactivation and gene repression (4). To determine if PTEN is responsible for regulating Akt phosphorylation, we depleted PTEN with siRNA and analyzed the level of Akt pS473. Consistent with the established signaling pathway, depleting PTEN in IMECs expressing exogenous Myc restored active Akt (pS473) (Figure 1D). Taken together, this data demonstrates that Myc induces PTEN to inhibit AKT activation, highlighting the biological consequence of a relationship that has only been alluded to previously.
To ensure we were not observing artifacts of overexpressing Myc beyond physiological levels, we monitored the amount of exogenous Myc in these systems. The level of Myc overexpression in IMECs is within the range observed in breast cancer cell lines, while the amount of Myc reconstituted in myc−/− fibroblasts is similar to that of endogenous Myc in parental myc+/+ fibroblasts (Fig S1A) (18). Other than PTEN, several other phosphatases have been reported to inactivate Akt via dephosphorylation but depletion had no effect on pAKT levels (Fig S1B, C).
It was reported previously that Akt suppresses the activity of the Ezh2 histone methyltransferase by phosphorylation on Serine 21 (pS21) (23). Given the previous link in Drosophila between Myc-mediated gene repression and Polycomb complexes (11), we decided to examine the status of this modification using phospho-specific antibodies. We observed a loss of phospho-Ezh2 (pS21) in all cells with exogenous Myc expression and suppression of pAkt (Figure 1C) along with an increase in total Ezh2, suggesting an increase in the pool of active Ezh2 upon increased Myc expression. Additionally we found that Myc stabilizes Ezh2, without altering its mRNA levels, and that this stability is negatively correlated with its phosphorylation at Serine 21 (Fig S2). To test more directly if PTEN activation was responsible for reduced pEzh2, we used two independent siRNAs to deplete PTEN in IMECs and observed a significant increase in phospho-Ezh2 upon loss of PTEN (Figure 1D).
The PI3K/Akt pathway is frequently hyperactivated in a variety of tumors and thus many drugs are available to inhibit it. We used the compound LY294002 that inactivates PI3K to further investigate the role of the PI3K/Akt pathway in Ezh2 phosphorylation (24). We found that inhibiting the PI3K/Akt pathway with LY294002 in parental IMECs prevents Akt activation and phosphorylation at Serine 473 (Figure 1E). In addition, loss of pS473 Akt induces loss of phospho-Ezh2 (pS21) and a modest increase in total Ezh2.
To test the consequence of activating Ezh2 through loss of phosphorylation, we assessed whole cell levels of the Ezh2-associated H3K27me3 modification that is indicative of repressed chromatin. Consistent with an increase in the active fraction of Ezh2, we observed Myc dependent elevation in global levels of H3K27me3 in both cell lines (Figure 2A). This genome-wide increase in H3K27me3 could contribute to the large number of genes that are Myc repressed. This led us to hypothesize that Myc mediated inhibition of the PI3K pathway via PTEN upregulation results in increased genome-wide gene repression elicited by PRC2. It is important to note that because we analyzed histones from whole cell extracts, we cannot distinguish between chromatin bound and non-chromatin bound histones.
To assess a role for Ezh2 activity and H3K27me3 in gene repression, we studied several Myc-repressed genes in two different cell systems. As discussed earlier, the first Myc repressed gene described was MYC itself so we tested if the autoregulatory feedback loop involved repression by H3K27me3. In IMECs, we analyzed the Myc-repressed genes MYC, SFRP1, DKK1, MYPN, ANKRD1 and HHIP (Figure 2B). It was previously shown that repression of SFRP1 and DKK1 is functionally important for Myc-mediated transformation (6). MYPN, ANKRD1 and HHIP are additional Myc repressed genes obtained from microarray data in IMECs (25). We observed an enrichment of the H3K27me3 modification at the endogenous human MYC promoter in response to autorepression by ectopic Myc (Figure 2C). A similar enrichment of H3K27me3 was also observed at exon 3 of MYC and at a region 30kb downstream of the MYC gene encompassing the transcriptional start site of the noncoding PVT RNA (Figure 2C and Figure S3B), which is regulated in parallel with MYC (26). These data suggest that the H3K27me3 mark may extend over a large area, consistent with spreading of H3K27me3 observed in Drosophila (27). As with the MYC gene itself, we found significant enrichment of the PRC2-mediated H3K27me3 modification in response to exogenous Myc expression at the promoters of all repressed genes tested (Figure 2C). In line with the propagation of H3K27me3 along gene bodies, we observed an enrichment of H3K27me3 at exon 2 of SFRP1, approximately 10 kb downstream of the promoter (Figure 2C). In contrast, two well-studied Myc activated target genes, nucleolin (NCL) and fibrillarin (FBL), did not show an enhancement of H3K27me3 in response to exogenous Myc (Fig. S3B), and Myc activated ribosomal protein targets were unaffected by overexpression of Ezh2 in myc−/− fibroblasts (Supplemental Table S1). We also found no binding of Myc protein itself at any of the Myc repressed promoters (Figure 2E).
Since Ezh2 is the methyltransferase responsible for the H3K27me3 mark, we analyzed genes with enriched H3K27me3 for Ezh2 binding by chromatin immunoprecipitation. As expected, Ezh2 was detected at repressed genes with the H3K27me3 mark (Figure 2D). Surprisingly, we did not observe enrichment of Ring1b, which is a component of PRC1, at Myc-repressed genes in IMEC cells (Figure S3A). This is consistent with a recent report showing the presence of PRC2, independent of PRC1, at bivalent domains in ES cells (28).
To further validate the role of PRC2 in Myc-mediated repression, we used the myc−/− rat fibroblast cell line in which two different selectable markers, neo and his, have been placed under control of the endogenous MYC promoters (29). Reconstituting Myc expression into these cells represses the endogenous MYC promoter similar to autorepression (30). We observed a strong induction of the H3K27me3 mark in response to reconstituted mouse Myc expression at the endogenous MYC promoter, at previously reported Myc repressed genes GADD45 and PDGFRb, and at additional Myc repressed genes SOD3, NCAM1 and EXPI which were selected from unpublished microarrays (Figures 2F and 2G). Additionally, we detected an accumulation of H3K27me3 on the HOXA1 promoter, a known Polycomb repressed target, but not at the promoter of a Myc activated target, Nol5a (Fig S3C). Ezh2 was also detected in these repressed genes in accordance with the IMEC data (Figure 2H). Thus, enhanced levels of H3K27me3 appear to be a general feature of Myc-repressed genes, including the MYC promoter itself.
To test if PTEN and Ezh2 are required for Myc-mediated repression, we depleted each with specific siRNAs. PTEN depletion dramatically reversed the repression of the Myc-repressed genes and prevented the accumulation of the H3K27me3 modification (Figure 3A–B and FigS4B). Similarly, depletion of Ezh2 (Figure 3C) restored expression of endogenous human MYC, SFRP1, HHIP, MYPN, ANKRD1 and DKK1 genes (Figure 3D and FigS4A) and induced a loss of H3K27me3 at these promoters (Figure 3E). Thus both PTEN and Ezh2 are required for sustained repression and accumulation of H3K27me3. Unfortunately, myc−/− fibroblasts do not permit transient transfection of siRNA to perform the same experiment.
To provide additional evidence that the PI3K pathway was linked to gene repression, we treated parental IMECs with LY294002. Chemical inhibition of the PI3K pathway (Figure 1D) resulted in repression of Myc repressed genes without altering the transcriptional activity of Myc induced targets (Figure 3F). Along with reduced expression, Myc repressed promoters showed an accumulation of the repressive H3K27me3 mark upon treatment with LY294002 (Figure 3G). Altogether, these data provide strong evidence for the importance of Myc-mediated inhibition of the PI3K/Akt pathway to initiate and maintain repression of genes.
To test more directly if the levels of PTEN and Ezh2 can suppress the MYC promoter analogous to autoregulation, we performed a transient assay with a MYC promoter-luciferase reporter. Ectopic expression of Myc, PTEN and Ezh2 all suppressed luciferase expression to a similar extent (Figure 3H, left panel). Depletion of Ezh2 with RNAi or expression of a constitutively active form of AKT (Myr-AKT) (31) is sufficient to alleviate repression of the MYC promoter (Figure 3H, right panel). These data support a model where Myc and PTEN exist in homeostatic balance to regulate Myc mediated gene repression and autoregulation, which is supported by a previous report that small changes in PTEN levels is sufficient to alter PI3K/Akt signaling and promote tumorigenesis (32).
While we observed a strong dependence on Ezh2 for Myc-mediated gene repression, neither cell system exhibited a similar dependence on Miz-1 (ZBTB17), a zinc finger transcription factor previously reported to play a role in Myc-mediated repression (33, 34) (Figure S4 C–G). We also did not observe Miz-1 occupancy at Myc-repressed gene promoters in IMECs (Figure S4C) or a gain of repression in control (IMEC:Vec) cells upon loss of Miz-1 (Figure S4E, left panel). Similar conclusions could be drawn from myc−/− fibroblasts stably expressing a Myc mutant (V394D) defective in Miz-1 binding (35). myc−/− fibroblasts stably expressing exogenous mouse WT-Myc or V394E were able to repress the transcription of genes similarly (Figures S4F and S4G). These data suggest an alternate mechanism of Myc-mediated repression that is Miz-1 independent.
Since overexpression of PTEN could affect multiple signaling pathways, we wanted to determine if altered activity of Ezh2 alone could account for Myc-mediated repression and autoregulation. To this end, we analyzed gene expression in myc−/− cell lines stably expressing Ezh2WT, Ezh2-S21A (phospho-defective) and Ezh2-S21E (phospho-mimetic), and compared it to repression in response to Myc over-expression. Exogenous Ezh2 expression was comparable to endogenous levels for all three constructs (Figure 4A), which can be resolved because exogenous Ezh2 has a tag that alters its size. Notably, expression of Ezh2-S21A in myc−/− cells is sufficient to repress the MYC promoter similar to autorepression, and similar repression was also observed for other Myc-repressed genes (Figure 4B). Expression of Ezh2-WT induced a modest repression, whereas the Ezh2-S21A mutant repressed as strongly as Myc. In contrast, there was no repression at all with the Ezh2-S21E mutant. Ezh2-WT enhanced H3K27me3 levels at all three genes (Figure 4C). The phospho-defective mutant (S21A) induced even higher H3K27me3 levels, which correlated with stronger repression, whereas the phospho-mimetic Ezh2-S21E induced barely detectable changes (Figure 4C). Additionally, we analyzed the level of a different histone modification (H3K4me3) which is associated with actively transcribed promoters. Interestingly, we found that overexpression of either Myc or any form of Ezh2 led to a complete loss of H3K4me3 at Myc-repressed promoters, confirming the established inverse relationship between these modifications (Figure 4D) (36). In addition to Myc WT and Ezh2, we also analyzed Myc mutants with defects in either the transactivation domain (ΔMB2) or DNA binding domain (ΔC). Neither mutant had any effect on H3K27me3 or H3K4me3 levels (Figures 4C and 4D), consistent with their defect in gene repression (Figure 4B).
We were particularly interested in determining the amount of Myc-mediated repression that can be accounted for by enhanced Ezh2 activity and elevated H3K27me3 levels. We analyzed whole genome expression by microarray using RNA from myc−/− fibroblasts expressing Myc-WT or Ezh2-S21A. We chose the hyperactive, phospho-defective form of Ezh2 to avoid phosphorylation of Ezh2-WT by active Akt (Figure 1C, lane 4) in myc/− cells. Comparable levels of H3K27me3 were present in myc+/+ parental fibroblasts, Myc-reconstituted myc−/− fibroblasts and myc−/− fibroblasts expressing ectopic Ezh2-S21A (Figure 4E). After normalizing to empty vector, we selected all genes that were repressed 2-fold or more by Myc WT and assessed their response to Ezh2-S21A. There were 1802 probes (1019 unique genes) repressed >2-fold by Myc, and 1724 probes (999 genes) repressed by Ezh2 S21A. Of the Myc-repressed genes, 814 probes (462 genes) overlapped with Ezh2-S21A repression, or a 45% overlap of repressed genes (Fig 4F). These data strongly support an integral role for Ezh2 in Myc-mediated gene repression and autoregulation.
Additionally, in our experimental setting, we observed that the transforming activity of Myc is directly tied to functional PRC2, and thus gene repression (Fig 4G). Depletion of Ezh2 significantly impairs the ability of IMEC cells expressing exogenous Myc to form colonies in soft agar. These data corroborate previous reports that have linked Myc-mediated gene repression to its ability to transform cells (5, 37). Altogether, these data provide evidence that Myc-mediated repression is an important component of Myc biology and its function as a potent oncogene.
This study presents a new model for the mechanism of Myc-mediated repression based on activation of PTEN transcription and the modulation of Ezh2 activity and repressive histone modifications. While complete loss of the PTEN tumor suppressor is common in cancer, even modest changes in PTEN levels are sufficient to promote oncogenic transformation (32, 38). In parallel, it has been established that incremental changes in Myc expression are a common driving force in cancer and can also contribute to inherited cancer predispositions (3, 15, 39, 40). Given these findings along with the data presented here, we propose that controlling the amount of Myc in a cell by autoregulation via modulation of PTEN expression is pivotal in maintaining the delicate balance of normal growth. Our data are consistent with recent studies showing that elevated PTEN suppresses MYC expression (38).
The regulation of Ezh2 by Myc occurs through the Akt pathway, which was previously shown to directly modify Ezh2 activity by phosphorylation (23). Suppression of Akt activity by PTEN reduces Ezh2 phosphorylation, increasing methyltransferase activity and simultaneously increasing Ezh2 protein levels by protein stabilization. Suppression of pAkt, pEzh2, and Ezh2 methyltransferase activity are dependent on both the MB2 and C-terminal DNA binding domains of Myc, consistent with previous mapping of Myc domains required for gene repression and autoregulation. The dependence on the Myc transactivation domain for gene repression stems from a requirement to induce the expression of PTEN. A recent study shows consistent activation of the Akt pathway in Burkitt’s lymphomas (41), but it is difficult to interpret these findings in relation to the model presented here because tumors may acquire complex mutational profiles that drive oncogenic growth independent of the MYC pathway or gene repression.
We show that the MYC gene itself as well as numerous Myc-repressed genes acquire high levels of H3K27me3 which is necessary and largely sufficient to suppress transcription (Figure 5). Repression by H3K27me3 involves a number of mechanisms such as the recruitment of PRC1 to the H3K27me3 deposited by PRC2 (42), but we find no enrichment of PRC1 components at Myc-repressed promoters. However, since PRC2 can be recruited to the histone modification that it creates (43), a small increase in Ezh2 activity could amplify the regional H3K27me3 modification through positive feedback at responsive genes (43, 44). Stable association of PRC2 complexes with repressed genes could also block the activating H3K4me3 modification as we observe. We show that Myc and Ezh2 share 45% of their repressed genes and that overexpression of Ezh2 alone can recapitulate Myc-mediated gene repression and autoregulation. These data suggest that the modulation of Ezh2 is a major mediator of Myc gene repression.
One aspect of our proposed mechanism that remains unclear is why particular promoters are responsive to Myc-mediated repression. No common motif has been associated with Myc-repressed genes other than the core initiator element in the promoter (33, 34, 45). One possibility is that certain promoters are poised to respond to a variation in H3K27me3 levels because they exist at a threshold in the balance between repressive and activating chromatin configurations. A small shift toward elevated repressive histone modifications could nucleate a localized expansion in repressive chromatin and downregulate gene expression. The balance of repressive and activating chromatin could be highly variable between cell types for individual genes, which could explain why Myc-repressed genes are so variable. An alternate possibility that is not mutually exclusive is that PRC2 complexes are guided to specific genes by noncoding RNAs that may vary in different cell types (46) or that may even be Myc induced. Nevertheless, our study presents a novel feedback pathway linking the potent tumor suppressor PTEN to MYC regulation and global changes in gene expression and chromatin modification.
We thank Carol Ringelberg for help with microarray data analysis. We thank members of the Cole lab for helpful discussions and feedback. We also thank Dr. M.C. Hung for reagents. This work was supported by a grant from the National Cancer Institute (CA055248).
The authors disclose no potential conflicts of interest